in

Linear regression models for continuous and binary data

By Morten Frydenberg

morten@biostat.au.dk

The participants should obtain a basic knowledge of linear and logistic regression methods as applied within health science.

The student should be able to:

Apply both linear normal and binary regression methods.

Confidently read and understand the output of a regression analysis

Understand and evaluate the assumptions behind the model.

Work with regression models that include interaction/effect modification.

Communicate the main results of a regression analysis and the assumptions behind these as a part of a paper.

Contents

Estimates, confidence intervals, tests and model validation.

Working with categorical explanatory variables and the significance of the choice of reference group.

Estimation and test for interaction/effect modification.

The use of regression models for prediction.

Things to consider in building the model.

A short introduction other models for the analysis of binary responses.

Format

Six days. The time will be divided equally between lectures and exercise using our laptop.

The statistical package Stata will be used at the exercises. It is required that the participant has a valid access to Stata (version 13 or newer) and that the software is updated.

Participants that are used to work in SAS, R or SPSS, and have a laptop with the software might are welcome.

Requirements

A knowledge of the basic biostatistiscal concepts and methods on a level comparable to the postgraduate course

Time schedule November 2017

November 6th 8th 10th and 20th, 22nd 24th.

All days 9 - 16 (one hour lunch break 12 to 13).

Copy of slides available from the course homepage.

Copies of Stata syntax and the data sets used in the lectures and the exercise can be found in electronic

form at course homepage.

Recommended additional reading:

Juul S, Frydenberg M.

It can be bought via the Gradplan at Metrika.se

Kirkwood BR og Sterne AC.

Teachers

Morten Frydenberg (course leader), Associated professor, Ph.D. Section of Biostatistics, Aarhus University

Morten Overgaard, Ph.D -student, Section of Biostatistics, Aarhus University

Homepage: